KorvaJoin the waitlist
Tech

On One Tech Site, Longer Articles Did Not Mean Longer Reads

One focused response of about 2,500 words kept visitors for a typical 84 seconds, nearly four times the site median.

AI image
csarko.sh · What keeps readers on a long technical article · Cyrus Sarkosh

Across the articles on the technical site run by Cyrus Sarkosh, word count had almost no link to how long visits lasted.

Over roughly a month ending in mid-November 2026, the typical time readers spent with a visible article page was 21.5 seconds. Word count had almost no link to time spent, with a rank correlation of about 0.01, a score from minus 1 to 1 where zero means longer pages did not get longer visits.

The best performer was a roughly 2,500-word response built around a single practical question. It kept visitors for a typical 84 seconds across 18 visits, nearly four times the site median.

Across the site, only 10 to 14% of reads both dug into the main text and made it to the finish. Of 184 visits, 168 never opened a second page.

Questions people ask

What does 21.5 seconds actually measure?

It measures engaged time, meaning seconds the page was visible in the browser tab. Sarkosh also used a stricter check for whether any main section stayed mid-screen for five seconds. That check began Oct. 23 and covered 145 reads, and only 37% passed it.

What does a correlation of 0.01 mean?

It means there was almost no relationship between word count and time spent. Longer articles did not reliably get longer visits, and shorter ones did not reliably get shorter visits.

Does this mean writers should keep everything short?

No. The best performer ran about 2,500 words. With only 18 visits for that one answer, the numbers cannot show what caused the longer time.

More in Tech